Keep attribution metadata checksums identical across mirrors
Browse files- .ms_upload_cache +1 -0
- CHANGELOG.md +8 -8
- COPYRIGHT.md +15 -15
- NOTICE.md +9 -9
- README.md +91 -91
- files_sha256.json +86 -86
- provenance.json +30 -30
.ms_upload_cache
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"version": 3, "repo_id": "Esmerubphire/FeiyuVoice-RVC", "files": {"CHANGELOG.md|1792972157.2851024|754": {"hash": "cc5a27e1d6bd325a616a073e141fe674e76efc0ed8e60e3497341497304f86fe", "size": 754, "status": "c"}, "COPYRIGHT.md|1792972157.2820897|1516": {"hash": "532e097be8a2c3a4a972f89198b9b16186e6cc938fa8e1ad0b70eadead230afd", "size": 1516, "status": "c"}, "README.md|1792972157.2830968|5118": {"hash": "5742de05a4202d5ab6e552129615b1171f3c9bdc034349c2db1c5fcfd5cbe4f6", "size": 5118, "status": "c"}, "NOTICE.md|1792972157.2841084|902": {"hash": "4b455807a5a9bae90255e0287ab71eff1b6dbbb3a31c49e86245bd29e60b44d5", "size": 902, "status": "c"}, "provenance.json|1792972157.2851024|1087": {"hash": "d9019eba4688c70e51be3d7d1a440c71551fba5db0f67690abeb3dacb7031d21", "size": 1087, "status": "c"}, "files_sha256.json|1792972157.3101203|2795": {"hash": "18c75916da8159b8f0e0383250db964b48b057349c11cb7896583eeae02f3816", "size": 2795, "status": "c"}, "CHANGELOG.md|1792972337.9096107|746": {"hash": "4d0b83f88eb78466e9c37d70dee107f3546e914636cd8a68c1f8444688256487", "size": 746, "status": "u"}, "COPYRIGHT.md|1792972337.910607|1501": {"hash": "662c301210ecb93bf89f37889b8b8b0525d4ec03c1a8373c876fb1e78185e87a", "size": 1501, "status": "u"}, "NOTICE.md|1792972337.910607|893": {"hash": "a3505c135c16925dec4792e3d7440b9ab23c15b2d9917f0ad2ceb396e7c676b1", "size": 893, "status": "u"}, "README.md|1792972337.9116063|5027": {"hash": "7988398d9e11bd4604e322fac707570a0e8d69bc5d1a2102da07e73f68688278", "size": 5027, "status": "u"}, "provenance.json|1792972337.9116063|1057": {"hash": "e25e35293319f654cb43d2b0c21d2517414f8f52a2ac6d02b66765a2498f863a", "size": 1057, "status": "u"}, "files_sha256.json|1792972337.9156072|2709": {"hash": "9060f8bed5445bcc04f26b46965af4bd31ecc7ad108cc08426434920dcdab647", "size": 2709, "status": "u"}}}
|
CHANGELOG.md
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
-
# FeiyuVoice-RVC v1 · 更新记录
|
| 2 |
-
|
| 3 |
-
## 2026-10-25 · 说明与署名更新
|
| 4 |
-
|
| 5 |
-
- 补充原始音频来源：[星落落_oi](https://space.bilibili.com/30584746/)；训练素材均已取得原 UP 主授权同意。
|
| 6 |
-
- 新增 COPYRIGHT.md，并同步 README、NOTICE 与 provenance 的来源及训练授权信息。
|
| 7 |
-
- 配套声音服务 v0.2.0 提供 CUDA／WSL ROCm 可选环境；**AMD 尚无实体显卡测试，属于待实机验收的实验适配**。安装方法见 [agent 指南](https://github.com/Esmerubphire/astrbot_plugin_feiyu_voice/blob/main/server/INSTALL_AGENT.md)。
|
| 8 |
-
- 本次不修改模型权重、索引、参考音频或推理代码，不新增训练，不改写既有发行标签；校验清单只更新说明元数据的记录。
|
|
|
|
| 1 |
+
# FeiyuVoice-RVC v1 · 更新记录
|
| 2 |
+
|
| 3 |
+
## 2026-10-25 · 说明与署名更新
|
| 4 |
+
|
| 5 |
+
- 补充原始音频来源：[星落落_oi](https://space.bilibili.com/30584746/)；训练素材均已取得原 UP 主授权同意。
|
| 6 |
+
- 新增 COPYRIGHT.md，并同步 README、NOTICE 与 provenance 的来源及训练授权信息。
|
| 7 |
+
- 配套声音服务 v0.2.0 提供 CUDA／WSL ROCm 可选环境；**AMD 尚无实体显卡测试，属于待实机验收的实验适配**。安装方法见 [agent 指南](https://github.com/Esmerubphire/astrbot_plugin_feiyu_voice/blob/main/server/INSTALL_AGENT.md)。
|
| 8 |
+
- 本次不修改模型权重、索引、参考音频或推理代码，不新增训练，不改写既有发行标签；校验清单只更新说明元数据的记录。
|
COPYRIGHT.md
CHANGED
|
@@ -1,15 +1,15 @@
|
|
| 1 |
-
# 原始音频素材与训练授权
|
| 2 |
-
|
| 3 |
-
肥鱼声库的原始音频素材来自 B 站 UP 主 [**星落落_oi**](https://space.bilibili.com/30584746/)。
|
| 4 |
-
|
| 5 |
-
**训练素材均已取得原 UP 主授权同意，用于本项目声库训练。** 感谢星落落_oi 对本项目的支持。
|
| 6 |
-
|
| 7 |
-
这份署名适用于肥鱼 TTS、肥鱼 RVC 及配套 GitHub 项目的素材来源说明。TTS 包中的可爱／温柔参考音频为现有肥鱼合成链路生成的参考，并非新增真人说话录音；原始声源的署名与授权在此一并说明。
|
| 8 |
-
|
| 9 |
-
本次更新补充素材来源与训练授权信息，没有新增训练、修改模型张量或公开训练录音。原始音频及其中作品的权利归各自权利人；模型、代码与第三方组件继续遵循各自的分发说明，具体见对应仓库的 `NOTICE.md` 和随附许可。
|
| 10 |
-
|
| 11 |
-
## English
|
| 12 |
-
|
| 13 |
-
The original audio source for the Feiyu voice models is Bilibili creator **星落落_oi** ([profile](https://space.bilibili.com/30584746/)). All training materials were used with the creator's authorization and consent for this project's voice-model training. Thank you for supporting the project.
|
| 14 |
-
|
| 15 |
-
This update records source attribution and training authorization. It does not change model tensors, redistribute the training recordings, or replace the existing distribution terms and third-party licenses. The TTS package's cute/gentle reference audio is synthetic audio from the existing Feiyu pipeline, not additional human speech recordings.
|
|
|
|
| 1 |
+
# 原始音频素材与训练授权
|
| 2 |
+
|
| 3 |
+
肥鱼声库的原始音频素材来自 B 站 UP 主 [**星落落_oi**](https://space.bilibili.com/30584746/)。
|
| 4 |
+
|
| 5 |
+
**训练素材均已取得原 UP 主授权同意，用于本项目声库训练。** 感谢星落落_oi 对本项目的支持。
|
| 6 |
+
|
| 7 |
+
这份署名适用于肥鱼 TTS、肥鱼 RVC 及配套 GitHub 项目的素材来源说明。TTS 包中的可爱／温柔参考音频为现有肥鱼合成链路生成的参考，并非新增真人说话录音；原始声源的署名与授权在此一并说明。
|
| 8 |
+
|
| 9 |
+
本次更新补充素材来源与训练授权信息，没有新增训练、修改模型张量或公开训练录音。原始音频及其中作品的权利归各自权利人；模型、代码与第三方组件继续遵循各自的分发说明，具体见对应仓库的 `NOTICE.md` 和随附许可。
|
| 10 |
+
|
| 11 |
+
## English
|
| 12 |
+
|
| 13 |
+
The original audio source for the Feiyu voice models is Bilibili creator **星落落_oi** ([profile](https://space.bilibili.com/30584746/)). All training materials were used with the creator's authorization and consent for this project's voice-model training. Thank you for supporting the project.
|
| 14 |
+
|
| 15 |
+
This update records source attribution and training authorization. It does not change model tensors, redistribute the training recordings, or replace the existing distribution terms and third-party licenses. The TTS package's cute/gentle reference audio is synthetic audio from the existing Feiyu pipeline, not additional human speech recordings.
|
NOTICE.md
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
-
# Distribution notice
|
| 2 |
-
|
| 3 |
-
This repository distributes the selected FeiyuVoice-RVC v1 inference assets and supporting tools at the owner's request. No blanket open-source or commercial-use license is granted for the model, retrieval index, character identity, source recordings, or locally authored material by this notice. Consult the publisher for reuse terms not otherwise specified.
|
| 4 |
-
|
| 5 |
-
Applio-derived runtime source retains its upstream MIT license, supplied separately. Dependencies retain their own licenses. Their licenses do not automatically extend to the model or recording-derived assets. No training audio is included.
|
| 6 |
-
|
| 7 |
-
## 原始音频素材与训练授权
|
| 8 |
-
|
| 9 |
-
原始音频素材来自 B 站 UP 主 [星落落_oi](https://space.bilibili.com/30584746/)。训练素材均已取得原 UP 主授权同意，用于本项目声库训练。完整署名与范围见 [COPYRIGHT.md](COPYRIGHT.md)。
|
|
|
|
| 1 |
+
# Distribution notice
|
| 2 |
+
|
| 3 |
+
This repository distributes the selected FeiyuVoice-RVC v1 inference assets and supporting tools at the owner's request. No blanket open-source or commercial-use license is granted for the model, retrieval index, character identity, source recordings, or locally authored material by this notice. Consult the publisher for reuse terms not otherwise specified.
|
| 4 |
+
|
| 5 |
+
Applio-derived runtime source retains its upstream MIT license, supplied separately. Dependencies retain their own licenses. Their licenses do not automatically extend to the model or recording-derived assets. No training audio is included.
|
| 6 |
+
|
| 7 |
+
## 原始音频素材与训练授权
|
| 8 |
+
|
| 9 |
+
原始音频素材来自 B 站 UP 主 [星落落_oi](https://space.bilibili.com/30584746/)。训练素材均已取得原 UP 主授权同意，用于本项目声库训练。完整署名与范围见 [COPYRIGHT.md](COPYRIGHT.md)。
|
README.md
CHANGED
|
@@ -1,91 +1,91 @@
|
|
| 1 |
-
---
|
| 2 |
-
language:
|
| 3 |
-
- zh
|
| 4 |
-
- en
|
| 5 |
-
pipeline_tag: audio-to-audio
|
| 6 |
-
tags:
|
| 7 |
-
- rvc
|
| 8 |
-
- voice-conversion
|
| 9 |
-
- singing-voice-conversion
|
| 10 |
-
- feiyu
|
| 11 |
-
license: other
|
| 12 |
-
license_name: feiyuvoice-distribution-notice
|
| 13 |
-
license_link: https://huggingface.co/Esmerubphire/FeiyuVoice-RVC/blob/main/NOTICE.md
|
| 14 |
-
---
|
| 15 |
-
# FeiyuVoice-RVC v1 · 肥鱼翻唱声库
|
| 16 |
-
|
| 17 |
-
肥鱼角色的 RVC 歌唱／语音音色转换声库。**输入已有的干净人声，转换成肥鱼音色。** 模型不是文字转语音，也不独立从 MIDI 生成歌声。
|
| 18 |
-
|
| 19 |
-
下载镜像：[Hugging Face](https://huggingface.co/Esmerubphire/FeiyuVoice-RVC) · [ModelScope](https://modelscope.cn/models/Esmerubphire/FeiyuVoice-RVC)。两个平台发布同一份权重、索引和工具。
|
| 20 |
-
|
| 21 |
-
AstrBot 插件与独立声音服务源码：[astrbot_plugin_feiyu_voice](https://github.com/Esmerubphire/astrbot_plugin_feiyu_voice)。已有的说话声库：[FeiyuVoice-CosyVoice3](https://huggingface.co/Esmerubphire/FeiyuVoice-CosyVoice3)。
|
| 22 |
-
|
| 23 |
-
## 原始音频素材与训练授权
|
| 24 |
-
|
| 25 |
-
原始音频素材来自 B 站 UP 主 [**星落落_oi**](https://space.bilibili.com/30584746/)。**训练素材均已取得原 UP 主授权同意，用于本项目声库训练。** 感谢原作者的支持。完整署名与范围见 [COPYRIGHT.md](COPYRIGHT.md)。
|
| 26 |
-
|
| 27 |
-
## 模型与参数
|
| 28 |
-
|
| 29 |
-
| 参数 | 设置 |
|
| 30 |
-
|---|---|
|
| 31 |
-
| 权重 | `FeiyuVoice-RVC-v1.pth` |
|
| 32 |
-
| 检索索引 | `FeiyuVoice-RVC-v1.index` |
|
| 33 |
-
| 架构 | RVC v2 / 40kHz / F0 / HiFi-GAN / 单歌手 |
|
| 34 |
-
| 内容编码器 | ContentVec 768 |
|
| 35 |
-
| 基频方法 | RMVPE |
|
| 36 |
-
| 移调 | 0；不自动改变音域 |
|
| 37 |
-
| Index Rate | 0.3 |
|
| 38 |
-
| Protect | 0.33 |
|
| 39 |
-
| Volume Envelope | 1.0 |
|
| 40 |
-
| Speaker / Seed | 0 / 42 |
|
| 41 |
-
|
| 42 |
-
完整SHA256见 `manifest.json` 和 `files_sha256.json`。v1为选定最终声库的正式对外命名；权重和索引没有重新训练、量化或改写。
|
| 43 |
-
|
| 44 |
-
## 最快使用：导入现有 RVC 软件
|
| 45 |
-
|
| 46 |
-
在兼容的 Applio／RVC 引擎导入上述 `.pth` 和 `.index`，按表中参数转换干净的 WAV／FLAC 人声。歌曲伴奏需要先单独分离，转换后再自行混音。直接导入权重即可换声，不需要安装 AstrBot。
|
| 47 |
-
|
| 48 |
-
## 可选：人工弱化换气
|
| 49 |
-
|
| 50 |
-
随包提供离线 `换气区间标记.html`：打开转换完成的 `rvc_native.wav`，试听并标记需要降低的换气区间，然后导出绑定该音频SHA的JSON。处理算法只在人工选区内降低音量，区间中部为−24dB，边缘采用余弦形dB渐变；不生成气声、不改音高、不做EQ、压缩或响度补偿。无区间时不应用。
|
| 51 |
-
|
| 52 |
-
安装工具依赖后运行：
|
| 53 |
-
|
| 54 |
-
```bash
|
| 55 |
-
python -m pip install -r requirements-tools.txt
|
| 56 |
-
python apply_breath.py --input rvc_native.wav --regions my_regions.json --output-dir new_output
|
| 57 |
-
```
|
| 58 |
-
|
| 59 |
-
原音频不覆盖，处理区外逐样本保持不变；保留采样率、声道和长度。`regions.example.json` 仅示范格式，不能套用于新歌。
|
| 60 |
-
|
| 61 |
-
## 固定运行时的独立入口
|
| 62 |
-
|
| 63 |
-
`run_rvc.py` 不依赖历史实验脚本，但需要已部署的 Applio GPU 环境。按照 [server/RUNTIME.md](https://github.com/Esmerubphire/astrbot_plugin_feiyu_voice/blob/main/server/RUNTIME.md) 准备指定运行时和兼容文件；通用 ContentVec／RMVPE 资产从上游下载，不包含在此声库中。
|
| 64 |
-
|
| 65 |
-
复制 `runtime.template.json` 为 `runtime.local.json`，填写自己的Python和Applio路径。Windows＋WSL可执行：
|
| 66 |
-
|
| 67 |
-
```powershell
|
| 68 |
-
.\Convert.ps1 -VerifyOnly
|
| 69 |
-
.\Convert.ps1 -InputAudio 'C:\audio\clean-vocal.wav'
|
| 70 |
-
```
|
| 71 |
-
|
| 72 |
-
Linux在已配置环境中执行：
|
| 73 |
-
|
| 74 |
-
```bash
|
| 75 |
-
python -B run_rvc.py --verify-only
|
| 76 |
-
python -B run_rvc.py --input clean-vocal.wav --output-dir new_output
|
| 77 |
-
```
|
| 78 |
-
|
| 79 |
-
核验包含文件SHA、固定运行时代码和实际载入的网络摘要。已在当前Windows＋WSL机器的隔离发布副本中验证；不同GPU、分段和第三方引擎不保证逐位一致。此仓库是模型与本地工具，**不是托管在线推理API或免安装EXE**。
|
| 80 |
-
|
| 81 |
-
## 范围与限制
|
| 82 |
-
|
| 83 |
-
当前选定声库完成2000次优化器更新。训练歌声、身份、音域和来源限制仍影响结果；中文／英文及不同音域质量需要按实际材料试听，不提供新的身份准确率或自然度评分。索引为匹配声库的检索特征，不包含在本页展示的训练音频中。
|
| 84 |
-
|
| 85 |
-
按谱唱歌需要额外的 SVS 引擎与已对齐参考，声音服务可执行“谱子＋歌词 → SVS → 本RVC声库”。仅下载本仓库不会自动获得SVS模型或私人参考素材。
|
| 86 |
-
|
| 87 |
-
未包含训练录音、原曲、实验试听、训练恢复状态、Python/CUDA环境或本机凭据。分发和复用说明见 [NOTICE.md](NOTICE.md)，上游代码许可单独保留。
|
| 88 |
-
|
| 89 |
-
## 可选 CUDA／WSL ROCm 服务环境
|
| 90 |
-
|
| 91 |
-
配套 [声音服务与 agent 安装指南](https://github.com/Esmerubphire/astrbot_plugin_feiyu_voice/blob/main/server/INSTALL_AGENT.md) 提供 CUDA 与 WSL ROCm 运行环境，复用本仓同一份权重和索引。AMD 实体显卡尚未测试，适配属于待实机验收的实验支持；不
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- zh
|
| 4 |
+
- en
|
| 5 |
+
pipeline_tag: audio-to-audio
|
| 6 |
+
tags:
|
| 7 |
+
- rvc
|
| 8 |
+
- voice-conversion
|
| 9 |
+
- singing-voice-conversion
|
| 10 |
+
- feiyu
|
| 11 |
+
license: other
|
| 12 |
+
license_name: feiyuvoice-distribution-notice
|
| 13 |
+
license_link: https://huggingface.co/Esmerubphire/FeiyuVoice-RVC/blob/main/NOTICE.md
|
| 14 |
+
---
|
| 15 |
+
# FeiyuVoice-RVC v1 · 肥鱼翻唱声库
|
| 16 |
+
|
| 17 |
+
肥鱼角色的 RVC 歌唱／语音音色转换声库。**输入已有的干净人声，转换成肥鱼音色。** 模型不是文字转语音，也不独立从 MIDI 生成歌声。
|
| 18 |
+
|
| 19 |
+
下载镜像：[Hugging Face](https://huggingface.co/Esmerubphire/FeiyuVoice-RVC) · [ModelScope](https://modelscope.cn/models/Esmerubphire/FeiyuVoice-RVC)。两个平台发布同一份权重、索引和工具。
|
| 20 |
+
|
| 21 |
+
AstrBot 插件与独立声音服务源码：[astrbot_plugin_feiyu_voice](https://github.com/Esmerubphire/astrbot_plugin_feiyu_voice)。已有的说话声库：[FeiyuVoice-CosyVoice3](https://huggingface.co/Esmerubphire/FeiyuVoice-CosyVoice3)。
|
| 22 |
+
|
| 23 |
+
## 原始音频素材与训练授权
|
| 24 |
+
|
| 25 |
+
原始音频素材来自 B 站 UP 主 [**星落落_oi**](https://space.bilibili.com/30584746/)。**训练素材均已取得原 UP 主授权同意，用于本项目声库训练。** 感谢原作者的支持。完整署名与范围见 [COPYRIGHT.md](COPYRIGHT.md)。
|
| 26 |
+
|
| 27 |
+
## 模型与参数
|
| 28 |
+
|
| 29 |
+
| 参数 | 设置 |
|
| 30 |
+
|---|---|
|
| 31 |
+
| 权重 | `FeiyuVoice-RVC-v1.pth` |
|
| 32 |
+
| 检索索引 | `FeiyuVoice-RVC-v1.index` |
|
| 33 |
+
| 架构 | RVC v2 / 40kHz / F0 / HiFi-GAN / 单歌手 |
|
| 34 |
+
| 内容编码器 | ContentVec 768 |
|
| 35 |
+
| 基频方法 | RMVPE |
|
| 36 |
+
| 移调 | 0；不自动改变音域 |
|
| 37 |
+
| Index Rate | 0.3 |
|
| 38 |
+
| Protect | 0.33 |
|
| 39 |
+
| Volume Envelope | 1.0 |
|
| 40 |
+
| Speaker / Seed | 0 / 42 |
|
| 41 |
+
|
| 42 |
+
完整SHA256见 `manifest.json` 和 `files_sha256.json`。v1为选定最终声库的正式对外命名；权重和索引没有重新训练、量化或改写。
|
| 43 |
+
|
| 44 |
+
## 最快使用：导入现有 RVC 软件
|
| 45 |
+
|
| 46 |
+
在兼容的 Applio／RVC 引擎导入上述 `.pth` 和 `.index`，按表中参数转换干净的 WAV／FLAC 人声。歌曲伴奏需要先单独分离，转换后再自行混音。直接导入权重即可换声，不需要安装 AstrBot。
|
| 47 |
+
|
| 48 |
+
## 可选：人工弱化换气
|
| 49 |
+
|
| 50 |
+
随包提供离线 `换气区间标记.html`：打开转换完成的 `rvc_native.wav`，试听并标记需要降低的换气区间，然后导出绑定该音频SHA的JSON。处理算法只在人工选区内降低音量，区间中部为−24dB，边缘采用余弦形dB渐变；不生成气声、不改音高、不做EQ、压缩或响度补偿。无区间时不应用。
|
| 51 |
+
|
| 52 |
+
安装工具依赖后运行：
|
| 53 |
+
|
| 54 |
+
```bash
|
| 55 |
+
python -m pip install -r requirements-tools.txt
|
| 56 |
+
python apply_breath.py --input rvc_native.wav --regions my_regions.json --output-dir new_output
|
| 57 |
+
```
|
| 58 |
+
|
| 59 |
+
原音频不覆盖，处理区外逐样本保持不变；保留采样率、声道和长度。`regions.example.json` 仅示范格式，不能套用于新歌。
|
| 60 |
+
|
| 61 |
+
## 固定运行时的独立入口
|
| 62 |
+
|
| 63 |
+
`run_rvc.py` 不依赖历史实验脚本，但需要已部署的 Applio GPU 环境。按照 [server/RUNTIME.md](https://github.com/Esmerubphire/astrbot_plugin_feiyu_voice/blob/main/server/RUNTIME.md) 准备指定运行时和兼容文件；通用 ContentVec／RMVPE 资产从上游下载，不包含在此声库中。
|
| 64 |
+
|
| 65 |
+
复制 `runtime.template.json` 为 `runtime.local.json`，填写自己的Python和Applio路径。Windows＋WSL可执行：
|
| 66 |
+
|
| 67 |
+
```powershell
|
| 68 |
+
.\Convert.ps1 -VerifyOnly
|
| 69 |
+
.\Convert.ps1 -InputAudio 'C:\audio\clean-vocal.wav'
|
| 70 |
+
```
|
| 71 |
+
|
| 72 |
+
Linux在已配置环境中执行：
|
| 73 |
+
|
| 74 |
+
```bash
|
| 75 |
+
python -B run_rvc.py --verify-only
|
| 76 |
+
python -B run_rvc.py --input clean-vocal.wav --output-dir new_output
|
| 77 |
+
```
|
| 78 |
+
|
| 79 |
+
核验包含文件SHA、固定运行时代码和实际载入的网络摘要。已在当前Windows＋WSL机器的隔离发布副本中验证；不同GPU、分段和第三方引擎不保证逐位一致。此仓库是模型与本地工具，**不是托管在线推理API或免安装EXE**。
|
| 80 |
+
|
| 81 |
+
## 范围与限制
|
| 82 |
+
|
| 83 |
+
当前选定声库完成2000次优化器更新。训练歌声、身份、音域和来源限制仍影响结果；中文／英文及不同音域质量需要按实际材料试听，不提供新的身份准确率或自然度评分。索引为匹配声库的检索特征，不包含在本页展示的训练音频中。
|
| 84 |
+
|
| 85 |
+
按谱唱歌需要额外的 SVS 引擎与已对齐参考，声音服务可执行“谱子＋歌词 → SVS → 本RVC声库”。仅下载本仓库不会自动获得SVS模型或私人参考素材。
|
| 86 |
+
|
| 87 |
+
未包含训练录音、原曲、实验试听、训练恢复状态、Python/CUDA环境或本机凭据。分发和复用说明见 [NOTICE.md](NOTICE.md)，上游代码许可单独保留。
|
| 88 |
+
|
| 89 |
+
## 可选 CUDA／WSL ROCm 服务环境
|
| 90 |
+
|
| 91 |
+
配套 [声音服务与 agent 安装指南](https://github.com/Esmerubphire/astrbot_plugin_feiyu_voice/blob/main/server/INSTALL_AGENT.md) 提供 CUDA 与 WSL ROCm 运行环境，复用本仓同一份权重和索引。AMD 实体显卡尚未测试，适配属于待实机验收的实验支持；不代表任意 A 卡或包内启动器已获得实机认证。版本记录见 [CHANGELOG.md](CHANGELOG.md)。
|
files_sha256.json
CHANGED
|
@@ -1,86 +1,86 @@
|
|
| 1 |
-
{
|
| 2 |
-
".gitignore": {
|
| 3 |
-
"bytes": 60,
|
| 4 |
-
"sha256": "a4b37cd4133f5d43b23608b07c7383b73b4a0f9e411ffbaa959c35f4014dfec1"
|
| 5 |
-
},
|
| 6 |
-
"Applio_CODE_LICENSE.txt": {
|
| 7 |
-
"bytes": 1088,
|
| 8 |
-
"sha256": "72c54bda40795cf382097befa7f63d8c38c6adf9ceb9861e12424fc048d0f18a"
|
| 9 |
-
},
|
| 10 |
-
"Apply-Breath.ps1": {
|
| 11 |
-
"bytes": 571,
|
| 12 |
-
"sha256": "a6db334bbf8f65749fcef93ea841ec39614e871ee1c0a1ced769b68436197d55"
|
| 13 |
-
},
|
| 14 |
-
"apply_breath.py": {
|
| 15 |
-
"bytes": 4937,
|
| 16 |
-
"sha256": "1a025bf83653430a559a318df38f512d7e4f7bfcc0cd59917a2848921309dce5"
|
| 17 |
-
},
|
| 18 |
-
"configuration.json": {
|
| 19 |
-
"bytes": 146,
|
| 20 |
-
"sha256": "b129ab0c3c4a46f056f24fb54e2dfb7fb7b93c1c24260a8be12a218c729cce3d"
|
| 21 |
-
},
|
| 22 |
-
"Convert.ps1": {
|
| 23 |
-
"bytes": 1267,
|
| 24 |
-
"sha256": "38e51fb020e40db937443069e3d10ba5d154046162b48096086e57bfca5d49ec"
|
| 25 |
-
},
|
| 26 |
-
"FeiyuVoice-RVC-v1.index": {
|
| 27 |
-
"bytes": 32004419,
|
| 28 |
-
"sha256": "3821de84cf7ad3e2be40475ee06c261da03be25b0583e571f568da2658a0732a"
|
| 29 |
-
},
|
| 30 |
-
"FeiyuVoice-RVC-v1.pth": {
|
| 31 |
-
"bytes": 55191150,
|
| 32 |
-
"sha256": "bcc3f4de74f87e102358db704c0fb75fc700fd8f44f491494287cc0b071a1dfe"
|
| 33 |
-
},
|
| 34 |
-
"manifest.json": {
|
| 35 |
-
"bytes": 1003,
|
| 36 |
-
"sha256": "bfc33a35de1f30448c459d2bd9f28afb16bac34d6df8709e3405fb741d4ce223"
|
| 37 |
-
},
|
| 38 |
-
"NOTICE.md": {
|
| 39 |
-
"bytes":
|
| 40 |
-
"sha256": "
|
| 41 |
-
},
|
| 42 |
-
"provenance.json": {
|
| 43 |
-
"bytes":
|
| 44 |
-
"sha256": "
|
| 45 |
-
},
|
| 46 |
-
"README.md": {
|
| 47 |
-
"bytes":
|
| 48 |
-
"sha256": "
|
| 49 |
-
},
|
| 50 |
-
"regions.example.json": {
|
| 51 |
-
"bytes": 199,
|
| 52 |
-
"sha256": "fefd88f260fcf977900f7c5deb0ed629f4696bfa93926e52fc39cee699c010a1"
|
| 53 |
-
},
|
| 54 |
-
"requirements-tools.txt": {
|
| 55 |
-
"bytes": 30,
|
| 56 |
-
"sha256": "542ee9880c585f398204d52f44b5df4c55a315274c7d1f861674646b43e99c95"
|
| 57 |
-
},
|
| 58 |
-
"runtime.template.json": {
|
| 59 |
-
"bytes": 373,
|
| 60 |
-
"sha256": "4845e7f11b258e2fc655cf6ee2fa0377ef3347bbbc76deff16542696a3af8afd"
|
| 61 |
-
},
|
| 62 |
-
"runtime_pins.json": {
|
| 63 |
-
"bytes": 2038,
|
| 64 |
-
"sha256": "7ee4e6aa07b08ea510502f3d9dfebff6db4cfc83bdd431470c7a8140c1573c8f"
|
| 65 |
-
},
|
| 66 |
-
"run_rvc.py": {
|
| 67 |
-
"bytes": 7922,
|
| 68 |
-
"sha256": "d25fdc5b7125190faefb934d3e1b2581989805a06035d4d71e2ad4c454b9151a"
|
| 69 |
-
},
|
| 70 |
-
"validation.json": {
|
| 71 |
-
"bytes": 1562,
|
| 72 |
-
"sha256": "57eb3a408ea596054f0fb8788d30eac90756d29415335799e687dcd4a75c6450"
|
| 73 |
-
},
|
| 74 |
-
"换气区间标记.html": {
|
| 75 |
-
"bytes": 8439,
|
| 76 |
-
"sha256": "e07aa82db04c8e04077845713612bfa8b8c43088b8fc50b97d84af1e9b5e4835"
|
| 77 |
-
},
|
| 78 |
-
"COPYRIGHT.md": {
|
| 79 |
-
"bytes":
|
| 80 |
-
"sha256": "
|
| 81 |
-
},
|
| 82 |
-
"CHANGELOG.md": {
|
| 83 |
-
"bytes":
|
| 84 |
-
"sha256": "
|
| 85 |
-
}
|
| 86 |
-
}
|
|
|
|
| 1 |
+
{
|
| 2 |
+
".gitignore": {
|
| 3 |
+
"bytes": 60,
|
| 4 |
+
"sha256": "a4b37cd4133f5d43b23608b07c7383b73b4a0f9e411ffbaa959c35f4014dfec1"
|
| 5 |
+
},
|
| 6 |
+
"Applio_CODE_LICENSE.txt": {
|
| 7 |
+
"bytes": 1088,
|
| 8 |
+
"sha256": "72c54bda40795cf382097befa7f63d8c38c6adf9ceb9861e12424fc048d0f18a"
|
| 9 |
+
},
|
| 10 |
+
"Apply-Breath.ps1": {
|
| 11 |
+
"bytes": 571,
|
| 12 |
+
"sha256": "a6db334bbf8f65749fcef93ea841ec39614e871ee1c0a1ced769b68436197d55"
|
| 13 |
+
},
|
| 14 |
+
"apply_breath.py": {
|
| 15 |
+
"bytes": 4937,
|
| 16 |
+
"sha256": "1a025bf83653430a559a318df38f512d7e4f7bfcc0cd59917a2848921309dce5"
|
| 17 |
+
},
|
| 18 |
+
"configuration.json": {
|
| 19 |
+
"bytes": 146,
|
| 20 |
+
"sha256": "b129ab0c3c4a46f056f24fb54e2dfb7fb7b93c1c24260a8be12a218c729cce3d"
|
| 21 |
+
},
|
| 22 |
+
"Convert.ps1": {
|
| 23 |
+
"bytes": 1267,
|
| 24 |
+
"sha256": "38e51fb020e40db937443069e3d10ba5d154046162b48096086e57bfca5d49ec"
|
| 25 |
+
},
|
| 26 |
+
"FeiyuVoice-RVC-v1.index": {
|
| 27 |
+
"bytes": 32004419,
|
| 28 |
+
"sha256": "3821de84cf7ad3e2be40475ee06c261da03be25b0583e571f568da2658a0732a"
|
| 29 |
+
},
|
| 30 |
+
"FeiyuVoice-RVC-v1.pth": {
|
| 31 |
+
"bytes": 55191150,
|
| 32 |
+
"sha256": "bcc3f4de74f87e102358db704c0fb75fc700fd8f44f491494287cc0b071a1dfe"
|
| 33 |
+
},
|
| 34 |
+
"manifest.json": {
|
| 35 |
+
"bytes": 1003,
|
| 36 |
+
"sha256": "bfc33a35de1f30448c459d2bd9f28afb16bac34d6df8709e3405fb741d4ce223"
|
| 37 |
+
},
|
| 38 |
+
"NOTICE.md": {
|
| 39 |
+
"bytes": 893,
|
| 40 |
+
"sha256": "a3505c135c16925dec4792e3d7440b9ab23c15b2d9917f0ad2ceb396e7c676b1"
|
| 41 |
+
},
|
| 42 |
+
"provenance.json": {
|
| 43 |
+
"bytes": 1057,
|
| 44 |
+
"sha256": "e25e35293319f654cb43d2b0c21d2517414f8f52a2ac6d02b66765a2498f863a"
|
| 45 |
+
},
|
| 46 |
+
"README.md": {
|
| 47 |
+
"bytes": 5027,
|
| 48 |
+
"sha256": "7988398d9e11bd4604e322fac707570a0e8d69bc5d1a2102da07e73f68688278"
|
| 49 |
+
},
|
| 50 |
+
"regions.example.json": {
|
| 51 |
+
"bytes": 199,
|
| 52 |
+
"sha256": "fefd88f260fcf977900f7c5deb0ed629f4696bfa93926e52fc39cee699c010a1"
|
| 53 |
+
},
|
| 54 |
+
"requirements-tools.txt": {
|
| 55 |
+
"bytes": 30,
|
| 56 |
+
"sha256": "542ee9880c585f398204d52f44b5df4c55a315274c7d1f861674646b43e99c95"
|
| 57 |
+
},
|
| 58 |
+
"runtime.template.json": {
|
| 59 |
+
"bytes": 373,
|
| 60 |
+
"sha256": "4845e7f11b258e2fc655cf6ee2fa0377ef3347bbbc76deff16542696a3af8afd"
|
| 61 |
+
},
|
| 62 |
+
"runtime_pins.json": {
|
| 63 |
+
"bytes": 2038,
|
| 64 |
+
"sha256": "7ee4e6aa07b08ea510502f3d9dfebff6db4cfc83bdd431470c7a8140c1573c8f"
|
| 65 |
+
},
|
| 66 |
+
"run_rvc.py": {
|
| 67 |
+
"bytes": 7922,
|
| 68 |
+
"sha256": "d25fdc5b7125190faefb934d3e1b2581989805a06035d4d71e2ad4c454b9151a"
|
| 69 |
+
},
|
| 70 |
+
"validation.json": {
|
| 71 |
+
"bytes": 1562,
|
| 72 |
+
"sha256": "57eb3a408ea596054f0fb8788d30eac90756d29415335799e687dcd4a75c6450"
|
| 73 |
+
},
|
| 74 |
+
"换气区间标记.html": {
|
| 75 |
+
"bytes": 8439,
|
| 76 |
+
"sha256": "e07aa82db04c8e04077845713612bfa8b8c43088b8fc50b97d84af1e9b5e4835"
|
| 77 |
+
},
|
| 78 |
+
"COPYRIGHT.md": {
|
| 79 |
+
"bytes": 1501,
|
| 80 |
+
"sha256": "662c301210ecb93bf89f37889b8b8b0525d4ec03c1a8373c876fb1e78185e87a"
|
| 81 |
+
},
|
| 82 |
+
"CHANGELOG.md": {
|
| 83 |
+
"bytes": 746,
|
| 84 |
+
"sha256": "4d0b83f88eb78466e9c37d70dee107f3546e914636cd8a68c1f8444688256487"
|
| 85 |
+
}
|
| 86 |
+
}
|
provenance.json
CHANGED
|
@@ -1,30 +1,30 @@
|
|
| 1 |
-
{
|
| 2 |
-
"release": "FeiyuVoice-RVC v1",
|
| 3 |
-
"inference_weights_unchanged": true,
|
| 4 |
-
"matching_index_unchanged": true,
|
| 5 |
-
"optimizer_updates": 2000,
|
| 6 |
-
"training_summary": "Selected v0.5 accepted singing corpus; rejected refinement branches excluded",
|
| 7 |
-
"included": "Inference model, matching retrieval index and optional manual breath attenuation tools",
|
| 8 |
-
"excluded": [
|
| 9 |
-
"training recordings",
|
| 10 |
-
"test songs",
|
| 11 |
-
"optimizer/checkpoint recovery states",
|
| 12 |
-
"credentials",
|
| 13 |
-
"local runtime configuration"
|
| 14 |
-
],
|
| 15 |
-
"new_training_updates": 0,
|
| 16 |
-
"publication_authorized_by_owner": true,
|
| 17 |
-
"source_attribution": {
|
| 18 |
-
"creator": "星落落_oi",
|
| 19 |
-
"platform": "Bilibili",
|
| 20 |
-
"profile_url": "https://space.bilibili.com/30584746/",
|
| 21 |
-
"uid": "30584746"
|
| 22 |
-
},
|
| 23 |
-
"training_authorization": {
|
| 24 |
-
"status": "authorized_for_training",
|
| 25 |
-
"statement": "训练素材均已取得原 UP 主授权同意，用于本项目声库训练。",
|
| 26 |
-
"confirmation_source": "publisher confirmation",
|
| 27 |
-
"recorded_date": "2026-10-25",
|
| 28 |
-
"training_audio_distributed": false
|
| 29 |
-
}
|
| 30 |
-
}
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"release": "FeiyuVoice-RVC v1",
|
| 3 |
+
"inference_weights_unchanged": true,
|
| 4 |
+
"matching_index_unchanged": true,
|
| 5 |
+
"optimizer_updates": 2000,
|
| 6 |
+
"training_summary": "Selected v0.5 accepted singing corpus; rejected refinement branches excluded",
|
| 7 |
+
"included": "Inference model, matching retrieval index and optional manual breath attenuation tools",
|
| 8 |
+
"excluded": [
|
| 9 |
+
"training recordings",
|
| 10 |
+
"test songs",
|
| 11 |
+
"optimizer/checkpoint recovery states",
|
| 12 |
+
"credentials",
|
| 13 |
+
"local runtime configuration"
|
| 14 |
+
],
|
| 15 |
+
"new_training_updates": 0,
|
| 16 |
+
"publication_authorized_by_owner": true,
|
| 17 |
+
"source_attribution": {
|
| 18 |
+
"creator": "星落落_oi",
|
| 19 |
+
"platform": "Bilibili",
|
| 20 |
+
"profile_url": "https://space.bilibili.com/30584746/",
|
| 21 |
+
"uid": "30584746"
|
| 22 |
+
},
|
| 23 |
+
"training_authorization": {
|
| 24 |
+
"status": "authorized_for_training",
|
| 25 |
+
"statement": "训练素材均已取得原 UP 主授权同意，用于本项目声库训练。",
|
| 26 |
+
"confirmation_source": "publisher confirmation",
|
| 27 |
+
"recorded_date": "2026-10-25",
|
| 28 |
+
"training_audio_distributed": false
|
| 29 |
+
}
|
| 30 |
+
}
|